Ponytail 这工具说白了就是把一个“话不多、只写一行就能跑”的资深程序员角色塞进你的 AI agent。Ponytail 5 是彻底重做的版本，原文标出 53% code、41% time、26% cost、45% tokens 这组数字，还说 98% 的风险逻辑会带测试，不用它是 68%。它在 Claude Code 里做基准测试，同一个 agent 带不带这个 skill 对比，跑了 39 个任务，包括真实 FastAPI 加 React 仓库，用 Opus 5.5，每个跑 5 次。盲测里 Ponytail 5 回复以 110 比 67 赢过上一版，六个安全任务涉及 SQL 注入、路径穿越、伪造令牌、限流。原文没写具体安装命令，只把它当 skill 用，适合 AI 编码代理少写代码、多覆盖风险逻辑。
查看英文原文
Ponytail He says nothing. He writes one line. It works. Ponytail 5: rebuilt from the ground up. 53% code · 41% time · 26% cost · 45% tokens And yet: 98% of risky logic ships with a test. Without Ponytail: 68%. Benchmarked in Claude Code, the same agent with and without the skill: 39 tasks including a real FastAPI + React repo, Opus 5.5, 5 runs each. Details . Español · 한국어 · 简体中文 · 日本語 Already built with Ponytail You know him. Long ponytail. Oval glasses. Has been at the company longer than the version control. You show him fifty lines; he looks at them, says nothing, and replaces them with one. Ponytail puts him inside your AI agent. Numbers Two things the chart does not show: in a blind comparison, Ponytail 5's replies beat the previous Ponytail's 110 to 67. And on the six security tasks (SQL injection, path traversal, forged tokens, rate limiting, malformed CSV rows, caching) it passed all 30 runs: less code, no less safe. Method, per task tables and limits: benchmarks/results/2026 10 07 agentic.md. The rule was never "fewest tokens." It is: write only what the task needs, and never cut validation, error handling, security, or accessibility. The code ends up small because it is necessary, not golfed. Lower cost and latency are a side effect. Before / after You ask for a date picker. Without Ponytail, the agent installs a date picker library or builds a whole calendar by hand: 335 lines. Ponytail 5 first looks at what is already there: the repo has an Input component, and every browser has a date picker. It puts the two together. 10 lines. More survivors in examples/. The review, rebuilt /ponytail review used to look only for code to cut. Now it reviews like the senior dev who gets paged when it breaks: it reads the code your change touches, not just the diff, and checks bugs, security, real load, missing tests, speed, and what to cut. Each finding says what the code does, what goes wrong, how to fix it, and what happens i